Incubator team: Could we have more translate TO French questions?
first of all, thanks for making this course so complete, useful and entertaining to use! I use it to refresh my age-old school French which I learned almost 20 years ago and it works great!
There is however one small thing that I feel could be improved: I think if we had more questions that ask you to translate an English phrase to French, and I mean free translation and not multiple choice, this would be more challenging and it would also force you to pay more attention to word gender, endings, accents etc. These small details are so easy to miss or ignore if you just write in a language you already know.
I appreciate that these questions would be a bit more challenging but to me that would be the best way to really go from passive to active vocabulary that you can use in an actual conversation.
It would be great if this course could put more emphasis on that.
Other than this, thanks again for all the brilliant work, it is much appreciated!
incubator teams can't do anything for that, only Duo staff can.
This is not something incubators teams have control on: teams define as many exercises in one way than the other.
Simply because, when adding a new exercise (=a sentence in the target language), contributors have to enter all possible translations into the "native" language and, then, to enter all translations back to target language of the answer you chose to be the best one (from target to "native" language).
The fact that there is more (easier) exercises from target to "native" language is due to Duolingo algorithm(s) that tries(-y) to optimize the exercises given to users in order to maximize the metrics Duo has set-up to evaluate what is the best way to teach each course.
So they can make changes so the proportion changes but it's not a "change a parameter" or "add more of those exercises" with immediate and precise effect thing.
Incubator teams (at least those I spoke with during the past year) all asked Duo to review things (about this point) too so there are more of those "translate to target language" exercises.
Hi, thanks for the quick and thorough answer! I could not find any way to contact duolingo directly, is there any way to contact whoever-it-is-one-should-talk-to (or did I just miss it)?
From what I understood it is not as easy as changing a value in some dashboard/control, but by changing a parameter (or several) in the algorithm is should indeed be possible? Maybe one could even pull this to the front end, so that end users could control this behaviour themselves (through some easy/medium/hard setting maybe?)
If this was sorted, I think I would call duolingo pretty darn perfect! Would be great to get an answer from their team!
*edit: just read your edit after sending, sounds like a start. Do you know of their answer?
Duo development tends to do thing the way they do because A/B tests show more engagement, faster learning, more efficient learning, etc. They are big on A/B tests and statistics rather than listening to what we think we want. I can see their point. A lot of times people misinterpret their own success or failure, or time spent learning, or whatever.
Many people like to take the "reverse course" - for you, that would be learning English for French speakers. Just turn off your mic and speakers so you aren't asked to "type what you hear" in English or practice your English pronunciation. Your exercises will be weighted so that you are translating from English into French (target into native) more.
"contact": maybe through the "bug report" button here. I don't see other way even if not perfect since not really a bug.
From what I understood it is not as easy as changing a value in some dashboard/control, but by changing a parameter (or several) in the algorithm is should indeed be possible?
Yes. Or define additional metrics that tends to push the other way than the current ones.
Maybe one could even pull this to the front end, so that end users could control this behaviour themselves
Duo works on statistical results (testing one thing against another and keeping the best) and what an learner think is best for her/his learning is not always what's in fact the bet/more efficient. So i'm not convinced it'd be the best way to do and do believe that it's not Duo vision.
Do you know of their answer?
That they agree and that it is in the TO DO list, that they are continuously working on improving current metrics and searching extra and/or better ones.
So, they are aware, know that it can be improved and looking for/testing improvements.
We just have to be patient. ;)
Great, sounds like something is on the way. Anyway, thanks for taking the time to answer in such depth!
As of last night, I have been presented with voice exercises that do precisely that. I was very surprised and pleased as the questions were fairly difficult and structured with the grammar that you see in the fourth leg of the tree. So, your request is rather timely. I'm sure this is a beta run so you may or may not have this. Once they perfect it then hopefully it will negate the necessity of doing the reverse tree.
I know this isn't the right place to get 'official' attention, but I absolutely agree that it would be great to see more 'into french' translation. It would make Duo tremendously more useful and challenging to me - Duolingo's a brilliant site, but I do feel it suffers from the lack of emphasis on 'blind' translation...
Is there an "immersion" page for French learners of English that would give you those exercises?
I completely agree that being asked to translate into your target language is a better way of learning, but surely it is very difficult to make a computerised system that can mark 'free translation'?
Not downvoting, but for me the benefit of Duolingo is the vocabulary and the beginnings of grammar. It is a decent suggestion though. There needs to be a mix.
I personally have issues remembering words properly when I only really get French to English instead of an even mix of English to French, so altering this would help a lot.